Thank you for Subscribing to Utilities Business Review Weekly Brief
Joseph Kao, CEOFor many municipal utilities and electric cooperatives, aging transformer fleets are becoming a growing operational risk. As replacement lead times stretch into years, they need continuous insight into transformer health to plan maintenance and capital investments and make informed replacement decisions well ahead of time. The question is whether they can keep the assets safely in service long enough for new equipment to arrive.
Magnefy empowers utilities to make transformer maintenance decisions with greater confidence with its new-generation predictive monitoring solution for the power industry.
Based on sensing technology innovations developed at Stanford University, Magnefy’s end-to-end platform continuously assesses transformer health and detects electrical anomalies months before failure. Combining proprietary high-speed electromagnetic sensing with advanced machine learning and AI, the platform captures high-fidelity electrical signals, accurately classifies fault types and converts complex operational data into clear, actionable maintenance recommendations.
Compact, non-invasive and affordable, the platform makes continuous, condition-based monitoring practical at scale for municipal utilities, electric cooperatives and renewable energy operators. It enables proactive maintenance decisions based on real-time operational intelligence that strengthens grid reliability and resilience.
“If the electric grid functions like a nervous system, transformers are its nerves. Our goal is to build the edge intelligence layer that continuously listens to those nerves, turning their signals into the insights utilities need to keep power flowing,” says Joseph Kao, CEO.
Listening to the Electrical Heartbeat of Transformers
For decades, utilities have relied on dissolved gas analysis (DGA) to assess transformer health, which depends on periodic oil sampling. By the time chemical indicators reveal insulation degradation, opportunities for early intervention may be lost. Even online DGA systems are expensive to deploy widely and often generate data that is difficult to translate into timely maintenance decisions.
Magnefy takes a different approach.
Its platform’s electromagnetic sensing system listens directly to the electrical “heartbeat” of the transformer. Delivering approximately ten times the sensitivity of conventional monitoring approaches, the technology enables operators to detect insulation degradation months earlier, creating valuable time to schedule maintenance before outages occur.
-
If the electric grid functions like a nervous system, transformers are its nerves. Our goal is to build the edge intelligence layer that continuously listens to those nerves, turning their signals into the insights utilities need to keep power flowing.
Magnefy’s sensing system installs on transformer ground connections without requiring equipment shutdowns. Once deployed, it continuously detects harmonics, partial discharges, corona activity, arcing and other electrical anomalies. It also identifies whether faults originate inside the transformer tank or within associated bushings, which aids maintenance planning.
The platform further combines proprietary sensing data with historical DGA records, temperature, pressure, vibration and other operational inputs to augment predictive monitoring.
Its machine learning and deep learning models accurately classify fault types while minimizing false positives. An AI copilot then converts these findings into IEEE-aligned maintenance recommendations, which operators can use to determine whether continued monitoring, further inspection or immediate intervention is required.
Behind the platform is a multidisciplinary team with expertise in advanced sensing, IoT, artificial intelligence and power engineering. The team is supported by Stanford advisors with extensive grid asset monitoring experience.
Expanding Intelligence across Critical Power Infrastructure
A number of utilities across North America are adopting the platform due to its advanced monitoring capabilities.
When Hyde Park Electric in Vermont unexpectedly lost nearly half the insulating oil in one of its critical transformers, it faced a challenge familiar to utilities. The aging asset was expected to remain operational for another five years, yet replacing it could take more than six.
Magnefy enabled Hyde Park Electric to gain continuous visibility into the asset's condition. This led the utility to make informed maintenance decisions that would keep the deteriorating transformer in service until a replacement could arrive.
As more utilities modernize their grids, plans are in place to bring the same sensing and AI capabilities to distribution transformers, renewable energy inverters, switchgear and power cables.
Magnefy is building a scalable intelligence layer for critical power infrastructure, which will continue to empower utilities to move from reacting to failures to anticipating and managing those months in advance.
Transformer Risk Intelligence for Constrained Utility Fleets
A transformer may still be running when a replacement is years away from delivery. For municipal utilities and cooperatives that cannot keep a large inventory of spares, knowing the condition of that transformer affects procurement as much as maintenance. Replacing it too early ties up capital unnecessarily. Missing a developing fault creates the opposite risk: an outage at a substation or generation site with no replacement readily available.
Periodic dissolved gas analysis provides useful information, but a lot can happen between samples. Electrical faults may develop during those intervals, and the chemical signs may not become clear until insulation damage has progressed. Online monitoring can provide more continuous visibility, although the cost may make it impractical to use across an entire mixed fleet. The information can also end up spread across different monitoring systems, leaving engineers to piece together readings when they are trying to judge fault severity or remaining service life.
Early detection is valuable only if engineers can trust what the system is telling them. An emerging internal fault has to be distinguished from normal background noise without every unusual waveform triggering an alarm. Sensitivity is only part of that equation. Maintenance teams also need to know what kind of problem they may be dealing with, whether it is partial discharge, arcing, a bushing issue or a power-quality disturbance. Frequent false positives consume inspection time and, over time, can make crews less confident in the alerts they receive.
How the equipment is installed can make a significant difference across an aging fleet. Monitoring systems that call for shutdowns, tank modifications, wiring changes or large numbers of sensors are harder to justify when transformers need to remain energized. Non-invasive monitoring avoids some of that disruption, but utilities still need to consider sensor coverage, supported voltage classes, calibration requirements and how an event is localized. Ideally, the same architecture can be deployed first on a troublesome transformer and later expanded across the fleet without having to redesign the monitoring program.
“Its continuous health assessment supports recommended maintenance actions and can incorporate historical dissolved gas records along with other condition data.”
The data also has to lead somewhere. Dynamic health scores, degradation trends, fault localization and estimates of time to failure can inform decisions about when an asset needs inspection and when replacement planning should begin. Electromagnetic signals may also need to be viewed alongside dissolved gas, pressure, temperature or vibration data to build a fuller picture of asset condition. Engineers need access to the technical detail behind those findings, while asset managers need enough clarity to decide whether continued monitoring is appropriate or a closer investigation is warranted.
The electrical environment can affect what the system sees. Transformers operating near inverters, for instance, may produce different noise patterns from standalone substation units. Continuous data collection can improve classification over time, but utilities still need confidence that the model can tell site-specific interference from fault signatures that appear consistently across assets. Physics-based interpretation adds another check on pattern recognition, particularly when one platform is expected to cover both low- and high-voltage equipment.
Magnefy combines high-speed electromagnetic sensors with machine-learning fault classification to provide this visibility without invasive installation. Its continuous health assessment supports recommended maintenance actions and can incorporate historical dissolved gas records along with other condition data. The system distinguishes partial discharge, arcing, corona and harmonic disturbance. Compact hardware, support for multiple sensors, installation without shutdown and coverage across voltage classes give utilities the option to start with selected assets and expand monitoring over time. For utilities managing aging transformers with long replacement lead times, the resulting insight can guide both maintenance timing and decisions about when to begin procurement.
...Read more
